1. Field of the Invention
The present invention relates to an ink jet recording apparatus which includes a recording means constituted by arranging a plurality of eject or ejection orifices, and performs recording by ejecting ink droplets onto a recording medium.
2. Related Background Art
With the widespread use of copying machines, information processors, such as wordprocessors and computers, and communication equipment, an apparatus for performing digital image recording using an ink jet recording head has rapidly spread as one of image forming (recording) apparatuses for these instruments. Such a recording apparatus employs a recording head (to be referred to as a multi-head hereinafter) constituted by integrally arranging a plurality of recording elements, and a head having integrated ink eject orifices and liquid channels is generally used as this recording head in order to improve the recording speed. The recording apparatus of this type generally includes a plurality of these multi-heads in order to perform color recording.
Unlike a monochromatic printer for printing only characters, an apparatus for performing color recording as described above must satisfy various factors such as color development properties, gradation characteristics, and uniformity. In particular, the uniformity is important because slight variations in nozzles produced due to differences between manufacturing processes of multi-heads affect the ink eject quantities or directions of the individual nozzles during printing, and this finally degrades an image quality as the density variation of a printed image.
This will be described in more detail with reference to FIGS. 65 and 66. Referring to (a) in FIG. 65, it is assumed that a multi-head 2001 is constituted by eight multi-nozzles 2002 for the sake of simplicity. As shown in (a) of FIG. 65, it is ideal that equal quantities of ink droplets 2003 are ejected in the same direction from the multi-nozzles 2002. If ink ejection is performed in this manner, dots of the same size are implanted on a paper surface as shown in (b) of FIG. 65, and a uniform image free from a density variation as a whole can be obtained ((c) in FIG. 65).
The individual nozzles, however, are actually different from each other as described above. Therefore, if printing is performed in the same manner as described above, variations are caused in the sizes and the directions of ink droplets ejected from these nozzles as shown in (a) of FIG. 66, and these ink droplets are implanted on a paper surface as shown in (b) of FIG. 66. As is apparent from (b) in FIG. 66, along the main scan direction of the head, blank portions which cannot satisfy an area factor of 100% are periodically present, the dots overlap each other thicker than needed, or a white stripe is formed as shown in the central portion of this drawing. The dots implanted in this condition form a density distribution shown in (c) of FIG. 66 with respect to the direction of the nozzle array. As a result, the above-mentioned phenomena are sensed as a density variation by human eye.
The following method is invented as a countermeasure against such a density variation. This method will be described below with reference to FIGS. 67 and 68A to 68C. According to this method, the multi-head 2001 is scanned three times in order to complete the print area shown in FIG. 65 or 66, and a half of this print area, i.e., a unit area of four pixels is completed by scanning the multi-head twice. In this case, the eight nozzles of the multi-head are grouped into four upper nozzles and four lower nozzles. In the first scan, each nozzle prints dots determined by thinning out given image data by about half in accordance with a predetermined image data array. In the second scan, image data dots of the remaining half are printed, thereby completing printing of the four-pixel unit area. This recording method will be referred to as a multi-pass recording method hereinafter.
The use of such a recording method can reduce by half the influences of the individual nozzles on a printed image even if a head having ejection characteristics, such as those of the multi-head shown in FIG. 66, is used. The resulting printed image is as shown in (b) of FIG. 67, in which black or white stripes are less conspicuous. Therefore, as shown in (c) of FIG. 67, the density variation is much more reduced than that shown in FIG. 66.
In performing recording of this type, image data is divisionally printed through the first and second scan operations, as two parts which complement each other, in accordance with a given predetermined array. This image data array (thinning pattern) is most often a checker pattern constituted by every other pixel in both the row and column directions, as shown in FIGS. 68A to 68C. Note that the recording methods of this type are described in the following patent specifications:
Japanese Laid-Open Patent Application No. 60-214670 PA0 Japanese Laid-Open Patent Application No. 60-214671 PA0 U.S. Pat. No. 4,622,561 PA0 U.S. Pat. No. 4,963,882 PA0 U.S. Pat. No. 4,967,203
In a unit print area (in this case, a four-pixel unit area), therefore, printing is completed through the first scan for printing a checker pattern and the second scan for printing a reverse checker pattern.
FIGS. 68A, 68B, and 68C are views for explaining the manner of completing recording of a predetermined area using the above checker and reverse checker patterns when a multi-head having eight nozzles is used as in the cases of FIGS. 65 to 67. In the first scan, recording of the checker pattern (hatched circles) is performed by using the four lower nozzles (FIG. 68A). In the second scan, paper feed is performed by the length of four pixels (half the head length), and recording of the reverse checker pattern (blank circles) is performed (FIG. 68B). In the third scan, paper feed is again performed by the length of four pixels (half the head length), and recording of the checker pattern is performed again (FIG. 68C). In this manner, paper feed in units of four pixels and alternate recording of the checker and reverse checker patterns are sequentially performed to Complete a four-pixel unit record area in each scan operation.
As described above, since printing is completed in each unit area by using two different groups of nozzles, a high-quality image free from a density variation can be obtained.
Even when this multi-pass recording is performed, however, the density variation is sometimes not at all eliminated depending on the print duty. In particular, another density variation is sometimes found in intermediate color portions or color-mixed portions. This phenomenon will now be described below.
Normally, image data of a given area to be recorded is already regularly arranged before reception by a printer. A printer stores a predetermined quantity of such data in a buffer and applies another mask (image array pattern), i.e., the checker or reverse checker pattern as described above to the data. When both the data array and the mask are set ON, printing of corresponding pixels is performed. FIGS. 69 to 71 are views for explaining the manner of this type of printing.
FIG. 69 illustrates already arranged data 2401 stored in a buffer, a checker pattern mask 2402 indicating pixels which can be printed in the first pass, a reverse checker pattern mask 2403 indicating pixels which can be printed in the second pass, and patterns 2404 and 2405 indicating pixels printed in the first and second passes, respectively.
Referring to FIG. 69, the buffer stores data already arranged to print 25% of a given area. In this data, in order to maintain a constant density in a designated predetermined area, an image to be printed is generally scattered as much as possible. The type of this image array depends on the area gradation method performed for the data during image processing before the data is transferred to a printer. The data 2401 shows an example of an image array for 25% data. When this data is printed through the masks 2402 and 2403, the data is equally divided and recorded in the first and second passes as represented by the pixel arrays 2404 and 2405, respectively.
If, however, data of exactly 50% is supplied as shown in FIG. 70, the array of data 2501 in which an image is scattered as much as possible coincides entirely with the array of either a checker pattern mask (2502) or a reverse checker pattern mask (2503), as can be seen easily from FIG. 70.
In this case, printing of the entire image data is completed in the first pass (2504), and no printing is performed in the second pass (2505). That is, all the print data (2501) on the same line is printed by the same nozzles. Therefore, since the influence of variations in nozzles is directly reflected on the density variation, the original purpose of the above divisional recording method cannot be achieved.
FIG. 71 shows the result of printing obtained when array image data having a higher duty than that shown in FIG. 69 or 70 is input. As shown in FIG. 71, a considerable difference is present between the numbers of printed pixels in the first and second passes. That is, although the density variation is reduced at high duties near 100%, it appears again with data having a low duty to a duty around 50%.
In addition, in the print area shown in FIGS. 68A to 68C, the checker pattern is implanted first in the upper half four-pixel unit area, and then the reverse checker pattern is implanted in this area. In the lower half four-pixel unit area, however, the reverse checker pattern and the checker pattern are implanted in this order. This phenomenon, combined with the problematic phenomenon described above, brings about another undesirable phenomenon in that a print area in which a large number of dots are implanted in the first pass while only a small number of dots are implanted in the second and a print area in which almost no printing is performed in the first pass while an enormously large number of dots are printed in the second alternately appear in units of half lengths of the head. This phenomenon introduces the following problem especially in color-mixed portions printed in accordance with the ink jet system.
That is, when solid printing of an intermediate color formed by properly ejecting inks of different colors to overlap and adjoin each other is performed, color variations regularly occur.
FIG. 72 shows a conventional head-divided printing method (to be referred to as an L/n-paper feed printing method hereinafter, where n indicates the number of divided areas).
In this method, a record area (L) of each recording head is divided into two parts, and each recording head records one of the checker and reverse checker patterns in the first scan as shown in FIG. 72. Thereafter, paper feed of a length of L/2 is performed, and a different nozzle group prints the other one of the checker and the reverse checker patterns, thereby completing the printing. Referring to FIG. 72, although eject orifice arrays cannot be actually seen, they are viewed from the above for convenience of description.
More specifically, in the first scan, the nozzles of a record area (1) of each recording head perform printing of an image to be recorded thinned out by half in the checker pattern. Paper feed is performed by a length of L/2. In the second scan, each recording head performs printing of the image thinned Out by half in the reverse checker pattern in both the record areas (1) and (2). As a result, printing in a portion corresponding to the record area (2) is completed. Paper feed of the length L/2 is performed again. In the third scan, the checker-pattern thinning printing is performed again by the entire record area, and so forth. In this manner, printing is repeatedly performed. Referring to FIG. 72, the contents in parentheses in the second and third scan operations indicate portions printed in the past.
The reason why an intermediate color variation is caused in this conventional example will now be described schematically below assuming that an eight-nozzle multi-head is used. In this case, four colors are cyan (C), magenta (M), yellow (Y), and black (K). Suppose, for example, that an intermediate color (yellowish green) overlapped at print duties of C 62.5% and Y 100% as shown in (a) of FIG. 73 is printed as image data to be recorded. When the intermediate color shown in (a) of FIG. 73 is printed using a checker pattern mask, printing of C and Y is performed at a 50% duty throughout the checker pattern mask in the first pass ((b) in FIG. 73). In the second pass, printing of the remaining duties, i.e., C 12.5% and Y 50% is performed ((c) in FIG. 73). The individual recording heads (of the respective colors) which form the images shown in (b) and (c) of FIG. 73 eject inks on pixels as indicated by (d), (e), (f), and (g) in FIG. 73.
FIG. 74 schematically shows the ejection position of the C recording head and the Y recording head in the first scan in accordance with the multi-pass printing method, and the resulting dots formed on a recording medium. Referring to FIG. 74, vertical stripe patterns indicate pixels where both the C and Y heads eject inks, and oblique stripe patterns indicate pixels where only the Y head ejects an ink. In the first scan, each recording head performs recording through the checker pattern mask by using the four nozzles of the record area (1). As a result, the checker pattern of dots where C and Y overlap each other is formed on the recording medium. Thereafter, paper feed of a length L/2 is performed to move the dots recorded in the first scan toward the record area (2).
FIG. 75 schematically shows an ejection position in the second scan and the resulting dots formed on the recording medium. In the second scan, the reverse checker pattern is printed by the entire record areas (1) and (2). As a result, the dots recorded in a paper portion corresponding to the record area (2) in this second scan overlap the checker pattern dots recorded in the first scan to complete the recording in this portion. Paper feed of the length L/2 is performed to move the paper portion corresponding to the record area (2) outside the record area and a paper portion corresponding to the record area (1) toward the record area (2).
It should be noted that when dots are printed to overlap other dots recorded previously, the dots implanted later tend to soak deeper in the direction of thickness of paper than the previously recorded dots at their overlap portions. FIG. 77 is a sectional view schematically showing this phenomenon. That is, a dye, such as a dyestuff, in an ejected ink bonds both physically and chemically with a recording medium. In this case, since the bond between the recording medium and the dye is finite, the bond between the dye of a previously ejected ink and the recording medium has highest priority as long as no large difference is present between the bonding forces of these dyes because of a difference between their types. Therefore, a large amount of the previously recorded ink dye remains on the surface of the recording medium, whereas the ink dye implanted later hardly bonds with the recording medium on its surface but soaks deep in the direction of thickness.
In FIG. 75, therefore, the dots recorded in the second scan are illustrated to overlap those recorded in the first scan from below.
FIG. 76 schematically shows an ejection position in the third scan and the resulting dots formed on the recording medium.
In the third scan, the checker pattern is printed by the entire record areas (1) and (2). As a result, the dots recorded in a paper portion corresponding to the record area (2) in this third scan overlap those recorded by the reverse checker pattern in the second scan to complete recording in this portion.
Although, however, the same quantities of inks are implanted in the portion currently corresponding to the record area (2) and the portion which is already moved outside the record area and in which printing is completed in the second scan, a difference is present between tones of color in the two portions to cause a color variation in these portions.
That is, since the checker pattern is recorded first in the portion outside the record area, a large number of pixels where both C and Y are ejected are present on the surface of the recording medium. In the portion corresponding to the record area (2), on the other hand, a large number of pixels where only Y is recorded are present on the surface of the recording medium. Therefore, yellowish green which is relatively strongly yellowish is formed in that portion.
A table of FIG. 78 explains this phenomenon using actual data. This table shows the results of printing of a halftone pattern consisting of a Y 100% duty and a C or K 50% duty performed in accordance with the multi-pass printing method. In this case, the printing was performed by both of a printing method (FINE1) in which Y 50% and C (K) 50% are recorded in the first pass and Y 50% and C (K) 0% are recorded in the second pass, and a printing method (FINE2) in which Y 50% and K 25% are recorded in both the first and second passes. (L*a*b) and a color difference in portions corresponding to the record areas (1) and (2) for performing different scan operations as described above were measured for two types of papers. According to these measurement results, an obvious difference is found between FINE1 and FINE2 even with the same print duty.
Because of the two problematic phenomena as described above, the conventional multi-pass printing method performed to compensate for nozzle variations can achieve only an image quality still unsatisfactory in terms of a density variation.
In addition, in the ink jet recording apparatus as described above, the ink eject orifices are grouped into a plurality of blocks and driven time-divisionally in units of these blocks, thereby preventing crosstalk. In this case, the crosstalk is a phenomenon that in a recording head mounting a plurality of ink eject orifices at a high density, ejection of an ink becomes unstable due to pressure waves or thermal diffusion from adjacent eject orifices.
FIGS. 79 and 80 are block diagrams showing conventional electric circuits for driving a printing head, and FIG. 81 is a timing chart showing signal waveforms of individual parts. In this case, assume that driving of a recording head for performing ink ejection is divided into four parts. A recording head 1 has 16 heaters 2, and eject orifices for ejecting an ink are provided in a one-to-one correspondence with the heaters 2. In this recording head 1, print data Si is set in a 16-bit shift register 5 by a print data sync clock CLKi, and a transistor array 3 is driven by switching on enable signals BEi1*, BEi2*, BEi3*, and BEi4*, thereby heating the heaters 2 in units of four heaters to perform printing. The signals BEi1*, BEi2*, BEi3*, and BEi4* are formed by a decoder 6. Note that a signal LATCH* is a control signal for latching the print data in a 16-bit latch 4, and a signal CARESi* is a reset signal for clearing the contents of the latch 4. A symbol * indicates that a corresponding signal is low-active.
The first heat operation is started by a heat start signal ((b) in FIG. 81), and a heat enable signal ((c) in FIG. 81) is kept HIGH during a print head driving period. The heat enable signal is formed by a flip-flop 7 which receives the heat start signal and a heat clock signal ((a) in FIG. 81). During the period in which the heat enable signal is HIGH, counters 8 and 9 are counted up by the heat clock signal. The value of the counter 8 causes a comparator 10 to generate a heat pulse* signal ((d) in FIG. 81) for driving the head 1. The comparator 10 keeps generating the heat pulse* signal until a pulse width set in a register 12 by a CPU (not shown) coincides with the count. When the count coincides with the set value, the comparator 10 switches the heat pulse* signal to LOW. The counter 9 is counted up for each cycle of the counter 8 to cause a comparator 11 to generate nozzle select1 and nozzle select2 signals ((e) and (f) in FIG. 81) for selecting divided blocks of the print head. The nozzle select1 and nozzle select2 signals select head-divide signals BEi1*, BEi2*, BEi3*, and BEi4* ((h) to (k) in FIG. 81) via the decoder 6. After switching the head-divide signals four times, the comparator 11 generates a heat end* signal ((g) in FIG. 81) and switches the heat enable signal to LOW, thereby ending the heat operation. In this circuit, all of the signals operate in synchronism with the heat clock signal. The heat enable signal indicating the heat period is started by the heat start signal and ended by the heat end* signal. The heat pulse* signal for actually heating the head 1 is divided into BEi1*, BEi2*, BEi3*, and BEi4* by divided block select signals, i.e., the nozzle select1 and nozzle select2 signals.
While the recording head is thus time-divisionally driven, divisional recording is performed.
FIGS. 82A to 82C show mask patterns for masking a signal to be recorded, in each of which hatched portions are portions for masking a signal and blank portions indicate recordable portions. Each mask is constituted by a 4.times.4 matrix, and an image signal is masked by repeating this mask pattern in both the vertical and horizontal directions. FIG. 83 is a view showing image recording processes, in which the eject orifice array of the recording head is equally divided into two parts (n=2). Referring to FIG. 83, a recording medium is fed in a direction A, and the recording head is scanned in a direction B.
First, data to be recorded is masked in accordance with a mask pattern (a) of FIG. 82A, and an upper-half image (a) of FIG. 83 is recorded by the first scan-recording. Subsequently, the recording medium is fed in the direction A by the length of an area (of eight nozzles in FIG. 83) obtained by equally dividing the eject orifice array (of 16 nozzles in FIG. 83) into two parts. The data is masked in accordance with a mask pattern (b) of FIG. 82A, and an image (b) of FIG. 83 is recorded by scanning the recording head 1 in the direction B. An image (c) of FIG. 83 is formed by these two recording processes. In the image (c), recording is completed in an area corresponding to the eight upper nozzles (upper-half nozzles) by performing the scan-recording twice using the eight upper nozzles and the eight lower nozzles of the recording head. Thereafter, these recording processes are repeatedly performed to form one image.
FIG. 84 is a view showing image recording processes in which the eject orifice array of the recording head is similarly equally divided into two parts and the mask patterns shown in FIG. 82B are used. The scan and feed operations of the recording head 1 and the recording medium are the same as those described above. Referring to FIG. 84, an image (a) is formed using a mask pattern (a) of FIG. 82B, and an image (b) is formed using a mask pattern (b) of FIG. 82B. An image (c) is formed through the two recording processes of the images (a) and (b).
FIG. 85 is a view showing recording processes performed when the eject orifice array of the recording head is equally divided into four parts (n=4). First, data to be recorded is masked in accordance with a mask pattern (a) of FIG. 82C, and an image (a) is recorded by the first scan-recording. Subsequently, the recording medium is fed in the direction A by the length of an area (of four nozzles in FIG. 85) obtained by equally dividing the eject orifice array (of 16 nozzles in FIG. 85) of the recording head 1 into four parts. The data is masked in accordance with a mask pattern (b) of FIG. 82C, and an image (b) is recorded by scanning the recording head 1 in the direction B. Thereafter, while the recording medium is similarly fed, an image (c) is formed using a mask pattern (c) of FIG. 82C, and an image (d) is formed using a mask pattern (d) of FIG. 82C. An image (e) is formed through these four recording processes. In an area of the image (e) corresponding to the four uppermost nozzles, recording is performed using different eject orifice groups obtained by equally dividing the eject orifice array into four parts, and an image is completely formed.
In each of the recording methods described above, it is possible to limit the quantity of an ink to be recorded on a recording medium at one time. Therefore, when recording is to be performed on a recording medium having a poor ink absorbency, especially when color recording is to be performed, a blur can be reduced to enable high-quality recording.
In the above recording methods, however, the heat signal of the head is divided, and the heat operation is time-divisionally performed in units of divided blocks. This results in a deviation between the positions of dots formed on the recording medium for reasons to be explained below.
FIGS. 88A and 88B are views showing the results of recording obtained when an image shown in FIG. 86 is recorded by the recording method shown in FIG. 83 while the eject orifices are divided as shown in FIG. 87A. FIG. 88A shows the result obtained when the head is not inclined, and FIG. 88B shows the result obtained when the head is inclined due to the mounting error or the like of the head. Note that dots 1, 2, and 3 are ejected by the first, second, and third scan operations, respectively, and each suffix indicates the number of a divided block.
A deviation shown in FIG. 88A is induced by a timing difference between divide signals BEi1, BEi2, BEi3, and BEi4, and this deviation L.sub.0 is given by L.sub.0 =.DELTA.T.upsilon. where .DELTA.T is the time difference between the heat signal generation timings of the divided blocks BEi1 and BEi4, and .upsilon. is the scan rate of the recording head. When the head is inclined as shown in FIG. 88B, the basic ejection position is varied by the time difference between the generation timings of the heat signals of the individual divided blocks and the inclination of the head, in accordance with whether the upper or lower portion of the head is used. As a result, the deviations satisfy a relation of L.sub.2 &lt;L.sub.1. In this case, recorded dots are scattered in an area corresponding to the eight upper nozzles, whereas they are concentrated in an area corresponding to the eight lower nozzles. Note that if the head 1 is inclined in a direction opposite to that shown in FIG. 88B, recorded dots are concentrated in the upper area while they are scattered in the lower area.
FIGS. 89A and 89B show the results of recording obtained when the image shown in FIG. 86 is recorded by the recording method shown in FIG. 83 while the eject orifices are divided as shown in FIG. 87B. FIG. 89A shows the result obtained when the head is not inclined, and FIG. 89B shows that obtained when the head is inclined. A deviation L.sub.0 in FIG. 89A is similarly given by L.sub.0 =.DELTA.T.upsilon. where .DELTA.T is the time difference between the generation timings of the heat signals of the divided blocks BEi1 and BEi4, and .upsilon. is the scan rate of the recording head. When the head is inclined as shown in FIG. 89B, the deviations satisfy a relation L.sub.2 &lt;L.sub.1 due to the time difference between the generation timings of the heat signals of the individual divided blocks and the variation in the basic ejection positions. As in the recording method described above, recorded dots are scattered in an area corresponding to the eight upper nozzles, whereas they are concentrated in an area corresponding to the eight lower nozzles.
The quality of an image recorded by this recording method is significantly degraded by stripe-like density variations produced by the above dense and sparse patterns of dots, which alternately appear in equally divided n areas (in this case, an area corresponding to the eight nozzles).
FIGS. 90A and 90B show images obtained by recording the image shown in FIG. 86 by the recording method shown in FIG. 84 when the eject orifices are divided as shown in FIG. 87B. FIG. 90A shows an image obtained when the head is not inclined, and FIG. 90B shows an image obtained when the head is inclined. When the head is inclined as shown in FIG. 90B, the deviations satisfy the relation L.sub.2 &lt;L.sub.1 as described above.
In this recording method, unlike in the above recording methods, no stripe-like density variations occur because dense and sparse dot patterns do not occur at equal intervals. However, since the difference between the deviations L.sub.1 and L.sub.2 is large, a texture which leads to degradation in image quality occurs.